Article Preview
Top1. Introduction
Heart arrhythmia is a category of heart disease which results from any disturbances in the rate, uniformity, and site origin or conduction of the cardiac electric impulse (Thaler, 1999). By examining and analyzing the combination of action impulse waveforms of the electrical signal of each heartbeat which are produced by various specialized cardiac tissues present in the heart, detection of abnormalities is possible (Luz et al., 2016). For the diagnosis of cardiac arrhythmia or any other heart rhythm disorder, electrocardiogram (ECG) is one of the most frequently used methods which helps in measuring and monitoring the impulses of the heart in the form of waveforms as it is non-invasive and efficient as well as an accurate method that provides useful information related to the heart impulse and other required details.
The muscles of the human heart contract and relax rhythmically for blood circulation throughout the body. Atrial sine node being the site of initial contractions act as a natural pacemaker and further pass through the other muscles. These electric signals propagate in a specific manner and follow a pattern (Herone and Smith, 2003), generating light electrical currents on the surface of the skin. These electrical signals are then measured with the help of electrodes and other appropriate devices. Electrodes measure the electric potential difference between the points and are amplified with optic isolation using suitable devices. Then these generated signals are passed through a high pass, followed by a low pass filter. At last, these signals are finally converted from analog signals to digital signals. This process of generating a graphical representation of digital signals done by the electrocardiogram (ECG) (Luz et al., 2016).
Recently, heart diseases have emerged as a major problem due to high rates of incidence and mortality. The treatment cost is another factor making it a serious disease as it requires long-term treatment and frequently entails expensive therapies for cure (Mozzafarian et al., 2015; National Center for Health Statistics, 2005). Every year, 17.34 million deaths are recorded owing to diseases that are related to the heart which in total accounts for 37% of the total deaths globally (Smith et al., 2005; Healthsquare, 2007; World Health Organization, 2014). Due to these reasons, with an expected progressive aging of the population globally, it may lead to an increased number of deaths from 17 million in 2016 to 24 million by 2030 (de Chazal et al., 2004). Many heart-related diseases can be diagnosed early via machine learning techniques (Anand et al., 2020; Das et al., 2020) and frequent itemset mining (Nayak et al., 2019). Several cases of arrhythmia appear as sequences of heartbeats with different timing or variation of ECG morphology. Classification of the correct type of heartbeat is an important step for the diagnosis of arrhythmia. The electric rhythm generated from the ECG signal is determined by analyzing and determining the consecutive heartbeats in the signal (Elhaj et al., 2016). According to the Association for the Advancement of Medical Instrumentation (AAMI), a heartbeat is further classified into five different categories which help in the diagnosis of arrhythmia (Association for the Advancement of Medical Instrumentation, 1998; Spach and Kootsey, 1983). Table 1 shows the categories of irregular heartbeats with their respective annotations.